32 In the following age group of individuals between 25-44, somewhat similar re- sults are observed: in most of the countries, the gap is in favour of women. The largest unconditional median gap, favouring men, is reported for Greece (0.309 log points), while in Norway there is an evidence of the largest gap in favour of women (0.395 log points). As for the age group of 45-65, the largest raw gap that favours men is observed in Portugal (0.415 log points). Moreover, there is a large number of countries where the unconditional gap benefits women more than men: Austria, Belgium, Czechia, Denmark, Germany, Greece, Hungary, Latvia, the Netherlands, Norway, Romania, Serbia, Slovakia, Slovenia, Sweden, and the UK. FIGURE 5: Median gap in public transfers for all countries. −.2 0 .2 .4 .6 Gap size <25 25−44 45−65 >65 Age group Austria −1 −.5 0 .5 Gap size <25 25−44 45−65 >65 Age group Belgium −.5 0 .5 1 Gap size <25 25−44 45−65 >65 Age group Bulgaria −1 −.5 0 .5 1 Gap size <25 25−44 45−65 >65 Age group Croatia −1 −.5 0 .5 1 1.5 Gap size <25 25−44 45−65 >65 Age group Czechia −1.5 −1 −.5 0 .5 Gap size <25 25−44 45−65 >65 Age group Denmark −1 −.5 0 .5 Gap size <25 25−44 45−65 >65 Age group Estonia −.2 0 .2 .4 Gap size <25 25−44 45−65 >65 Age group Finland −.5 0 .5 1 Gap size <25 25−44 45−65 >65 Age group France −.4 −.2 0 .2 .4 .6 Gap size <25 25−44 45−65 >65 Age group Germany −1 −.5 0 .5 1 1.5 Gap size <25 25−44 45−65 >65 Age group Greece −1 −.5 0 .5 Gap size <25 25−44 45−65 >65 Age group Hungary −.5 0 .5 1 Gap size <25 25−44 45−65 >65 Age group Latvia −1 0 1 2 Gap size <25 25−44 45−65 >65 Age group Lithuania −1 −.5 0 .5 1 Gap size <25 25−44 45−65 >65 Age group Netherlands −1.5 −1 −.5 0 .5 Gap size <25 25−44 45−65 >65 Age group Norway −.5 0 .5 1 1.5 Gap size <25 25−44 45−65 >65 Age group Poland −1 −.5 0 .5 Gap size <25 25−44 45−65 >65 Age group Portugal −2−1.5−1 −.5 0 .5 Gap size <25 25−44 45−65 >65 Age group Romania −.1 0 .1 .2 .3 .4 Gap size <25 25−44 45−65 >65 Age group Serbia −1.5 −1 −.5 0 .5 Gap size <25 25−44 45−65 >65 Age group Slovakia −.8 −.6 −.4 −.2 0 .2 Gap size <25 25−44 45−65 >65 Age group Slovenia −.5 0 .5 Gap size <25 25−44 45−65 >65 Age group Spain −1 −.5 0 .5 Gap size <25 25−44 45−65 >65 Age group Sweden −1 −.5 0 .5 1 Gap size <25 25−44 45−65 >65 Age group UK Raw gap Unexplained gap Note: confidence intervals are shown for 90% confidence level. Source: author’s calculation from EU-SILC 2016. Unlike from the results for all previous age groups, among the oldest, both un- conditional and conditional medium gaps in public transfers are in favour of men. The only exception is Denmark, where the unconditional gap favours women (0.044 log points). Moreover, all estimates are statistically significant. The largest uncondi- tional gap is observed in Austria, where the differences in observed characteristics can account for 0.116 out of 0.573 log points. In Estonia there is an evidence of the smallest gap in favour of men (0.023 log points). In the vast majority of countries both explained and unexplained part of the total gap are in favour of men (the ex- plained gap favours women only in Denmark, Slovakia, and the UK). 5. Results 33 Figure 5 shows that for most of the countries the unconditional median gap favouring men peaks at the oldest age group. One possible explanation for this observation could be high rate of tobacco and alcohol consumption among men (EHIS (2015) 18, Mäkelä et al. (2006)) throughout their life that may result in in- creased sickness and disability benefits at older ages. On the contrary, the short life-expectancy of males can explain why women have “advantage” at early years, which contributes to increased survivor’s benefits that is granted to people under retirement age, when a spouse or partner dies. Also, given the fact of increased women’s participation in education, this can contribute to higher educational al- lowances compared to men. 5.5 Impact of the institutional factors on the unexplained total income gap. The paper also studies the effects of institutional factors on the unexplained gap at different quantiles of the distribution. In total, eight institutional factors are an- alyzed: (1) union density, (2) strictness of employment protection legislation, (3) Kaitz index19, (4) maternity pay entitlement, (5) formal child care for children under 3, (6) pensions system design, (7) minimum wage setting, and (8) national minimum wage. The data on these factors has been collected from several sources20. The infor- mation on union density is collected from administrative data for year 2015 21. The employment protection indicator is collected for 2013, however, 2014 data is used for Slovenia and Lithuania, due to its absence for 2013. Kaitz index is also calculated for wages in 2013. As suggest by Christofides et al. (2013), maternity pay entitlement has been calculated as a product of maternity leave length (weeks) and pay rate (%). Information on formal child care is based on EU-SILC 2016 data and depicts average weekly hours spent at pre-school and day care centers. Pension design is ordinal variable with three levels: public, combined, and private. Likewise, national mini- mum wage and minimum wage setting variables are ordinal variables with 3 and 6 levels, respectively22. Table 6 reports Spearman correlation coefficients between the institutional fac- tors and unconditional total income gaps at the 20 th, 50th, and 90th quantiles across all age groups. As observed, union density has significant negative impact on the unexplained gap for individuals below age 25 and between 25-44. Blau and Kahn (1992, 1996) provide the evidence of reducing wage dispersion as unionization rates increase in a country. Given the fact the employment income is a large part of the total income for the individuals in the above-mentioned age groups, it is logical to expect the higher unionization rates to reduce the unexplained and, therefore, raw 18European health interview survey (EHIS), 2015. 19Kaitz index is defined as the ratio of nominal minimum wage to average wage. 20Data on (1), (2), (3), (4), (5), and (6) factors have been collected from OECD databases, while infor- mation on (7) and (8) variables are provided by the ICTWSS (Database on Institutional Characteristics of Trade Unions, Wage Setting, State Intervention and Social Pacts in 51 countries between 1960 and 2014). 21Estonia is the only country for which survey data has been used. Due to data limitations, 2014 and 2013 data have been used for Poland and Greece, respectively. 22National minimum wage levels are no statutory minimum wage, statutory minimum wage in some sectors, and statutory national minimum wage. Levels for minimum wage setting range from between agents collective agreement to the minimum wage set by government without a fixed rule. 34 total income gap. Another interesting finding is that the union density has no sig- nificant impact on high-income individuals. This is rather intuitive because high- income individuals are less likely to join trade unions and if so, bargaining power of trade unions in the negotiations for pay increase is expected to be low. TABLE 6: Correlation of unexplained total income gap with institu- tional factors. Age-group Percentile Union density Employee protection Kaitz index1 Maternity pay entitlement2 Formal child-care for children under 33 Minimum wage setting National minimum wage Pension design 20 0.197 0.121 -0.054 -0.344 -0.292 -0.014 -0.030 -0.378 < 25 50 -0.390* 0.139 0.285 0.062 0.103 0.507** 0.304 -0.416* 90 -0.211 -0.233 0.175 -0.013 0.114 0.061 0.232 0.069 20 -0.546** 0.012 0.498** 0.419* 0.064 0.815*** 0.616*** -0.295 25 −44 50 -0.493** -0.007 -0.032 0.746*** -0.108 0.390* 0.168 -0.240 90 -0.358 -0.100 -0.296 0.150 0.043 -0.066 -0.146 0.026 20 -0.040 0.230 0.612** 0.144 -0.153 0.421* 0.378 -0.359 45 −65 50 -0.191 0.204 0.233 0.387 -0.459** 0.276 0.309 -0.304 90 -0.118 -0.263 -0.129 -0.254 0.166 -0.170 -0.058 0.587*** 20 0.274 0.291 0.075 -0.221 -0.398* -0.164 -0.017 -0.255
65 50 0.233 0.237 0.064 -0.100 -0.508** -0.150 -0.085 -0.328 90 0.065 0.085 0.099 -0.143 -0.440* -0.101 -0.047 -0.232 1 Minimum wage divided by average wage. 2 Product of maternity leave length (weeks) and pay rate (%). 3 Average usual weekly hours for children using early childhood education and care services. Note: * p < 0.10, ** p < 0.05, *** p < 0.01 Source: author’s calculations. The employment protection is found to have no significant impact on the unex- plained total income gap. Also, the impact of pension design among the youngest individuals remains puzzling. However, Kaitz index, minimum wage setting, and national minimum wage have significant positive impact on the unexplained gap for individuals between 25-44 and 45-65. On the contrary, the minimum wage set- ting and national minimum wage have negative impact on the employment income gap for low income individuals (not reported in the paper). This result is in line with earlier literature, where it is shown that the introduction of minimum wage reduces the gender wage gap (e.g. see Bargain et al. (2018)). This could partially explain why the setting of minimum wage increases the unexplained total income gap: once minimum wages are set, those individuals who might be receiving some
- Conclusions 35 public transfers due to low income, they do not qualify for those benefits anymore, which might have been larger than the marginal increase in the employment income. Maternity pay entitlement is found to have significant negative impact on the unconditional total income gap for individuals between 25-44. Given the fact that women are most fertile in their twenties and early thirties, when the maternity leave and its pay rate are regulated by state laws, the increase of either of these indicators would put more burden on employer. Therefore, employers are more likely to offer lower wages to women. Also, Farré (2016) highlighted that more generous mater- nity leave increases women’s labor force participation in low paying jobs and their absence in high-profile occupations. The formal child-care has negative impact on the unexplained total income gap for the oldest individuals. It turns out to have significant impact not only on low- income but, also, on high-income individuals. In many countries informal child- care still dominates over formal care, especially for Eastern and Central European countries (Mills et al. (2013)). The burden of informal child-care, which is typically not paid, is usually put on grandmothers. The increase of formal child-care allows grandmothers to dedicate their time to paid tasks, which would otherwise be spent on free child-care for their grandchildren. 6 Conclusions The paper studies the gender income gap across 25 countries over 4 age groups. For the analysis survey data of European Union Statistics on Income and Living Con- ditions (2016) is used, which, in addition to other demographic and socio-economic information, collects data on personal and household level income. To examine the gender income gap, 17 different income sources are aggregated: 10 personal level and 7 household level. For extending the scope of the paper, these income sources have further been grouped into 3 categories: employment income, private transfers and capital income, and public transfers. To study the factors contributing to the gap, unconditional quantile regression is applied for both genders, and then the dif- ference is estimated by Oaxaca-Blinder decomposition method. The analysis showed that the gender income gap is still inarguable problem in EU. However, even though the gap is mostly in favour of men, this is not always the case. The reported results unveiled the the high level of heterogeneity among coun- tries. In addition, there is a large degree of variation observed within each country’s age group. In each country the conditional median gap demonstrates several pat- terns over these age categories. In Bulgaria and France, the unexplained median gap has U-shaped pattern In particular, the unexplained median gaps for age groups 25- 44 and 45-65 are lower compared to the other two groups. However, the conditional median gap has predominantly inverse U-shaped form in Belgium, Denmark, Esto- nia, Finland, Greece, Hungary, Latvia, Lithuania, the Netherlands, Portugal, Serbia, and Spain, where the unexplained gaps peak at age groups 25-44 or 45-65. Among the individuals below age 25, the largest unconditional median income gap in favour of men is reported for Bulgaria, while the largest unconditional me- dian income gap favouring women is observed in Greece. The intuition behind the